Everything we build is grounded in research on how oral assessment reveals understanding — and how it doesn't. We publish our findings openly and hold ourselves accountable to the evidence.
Does oral assessment surface understanding the essay alone hides?
Are AI scores within the same range as a panel of experienced teachers?
Do conversation scores differ across language background, disability, or socioeconomic indicators?
Every rubric, every prompt, every model decision — open to partner schools.
Working Papers
How conversation response patterns differ between students who submit AI-generated work and students who wrote their own — across secondary and tertiary institutions. We document the signals our scoring picks up on, the rates we observe, and where the method breaks down. Submitted to partner schools for review.
A grounded study of how teachers evaluate authenticity in student speech. We map the behavioural signals teachers use — vocabulary register shifts, hedging, over-specificity — into a trainable rubric validated against expert human raters.
A follow-up on students who completed Viva conversations across multiple assignment cycles. Early evidence suggests that knowing an oral conversation is coming changes how students engage with source material before submission.
How We Work
We document every scoring rubric, every prompt, every model decision in our methodology appendix — available to partner schools on request.
AI scores are calibrated against a panel of experienced teachers and updated when teacher-AI disagreement exceeds thresholds.
We actively test for score disparities across language background, disability status, and socioeconomic indicators. If we find them, we investigate them.
Viva scores are advisory signals. Final grade decisions remain with the teacher. We are explicit about this in every interface.
Partner with our research team
We collaborate with universities, school districts, and independent researchers working on assessment, integrity, and AI in education.
Reach out →